From the Washington Post

As calculated by federal statisticians, the productivity growth of U.S. factories has seemed quite impressive. Between 1991 and 2011, productivity more than doubled, meaning that a single worker today produces what two did 20 years ago, according to Bureau of Labor Statistics figures.

Except that it doesn’t mean this. And, unless there is just some weird co-incidence, it never has and never will mean this.

It means the ratio of output-to-workers has doubled, which could be achieved by lots of means – not least of which is simply changes in the composition of output.

So, if people were consuming  mostly processed food which required a lot of labor per output, but over time the consumption share moved to laptops which require much less labor per unit of output then productivity would rise, even though nothing at all has happened to any manufacturing process.

Indeed, this is possible even if labor productivity in each sector actually falls. Also, the reverse is true, labor productivity can fall even if the productivity in each sector rises.

This is in addition to the fact exploiting comparative advantage to ship US jobs overseas raising productivity and that innovation that raises the value of the output raises productivity.

Likewise if a cashier at a jewelry store rings up one $10,000 item an hour then he is more productive than a cashier at a grocery store who rings up 1000 $5 items an hour. This is despite the fact that by all appearances the cashier at the jewelry store is accomplishing less in the course of an hour.

This is because the statistics productivity is just not congruent with the common notion of being productive.

The piece goes on to discuss import price bias which would underestimate the total value of imported products

Critically, Houseman and others have shown that the price savings that U.S. factories have realized from outsourcing have incorrectly shown up as gains in U.S. output and productivity.

. . .

The federal statistical agencies, which have helped fund Houseman’s work, agree that the bias exists, though they say there might be other problems that are off­setting.

I don’t know for sure if this is what the agency heads were trying to communicate, but an important point is that while import price bias can produce granular statistics that do not measure what they claim to measure, they must do that at the expense of something else happening.

So, for example, if proper accounting shows that the real value of an IPad is 70% foreign rather than 50% foreign, then something has got to go the other way to explain how 50% of the revenue shows up in Apple’s bank account after supply chain costs.

Its not immediately clear what that thing is, but it has to be somewhere because by design these numbers all have to add up consistently.

Ultimately I really think the problem is that increasing productivity sounds like something that should make you feel good, but some of the results of the process  make most people feel bad.

And, so the argument is really over: should we feel good or bad about this evolution of events.